C Quigley
C Quigley
C Quigley
Callum Quigley
April 14, 2003
1 Introduction
We know the Universe to be governed by four fundamental interactions:
namely, the strong and the weak nuclear forces, electromagnetism and grav-
itation. It is a driving concept to unify these forces into a single, compre-
hensive theory. Though this task is far from its completion, there has been
much progress.
The first great landmark in its development is attributable to James
Maxwell, who in 1864 brought together the seemingly unrelated concepts of
electricity, magnetism and optics into the now well known theory of electro-
magnetism. Nearly a hundred years later, the weak force too, was combined
with the electromagnetic by Glashow, Weinberg and Salam, giving rise to
the electroweak theory. Currently, attempts are being made to find a Grand
Unified Theory which would explain all the forces, except gravity, as man-
ifestations of the same fundamental interaction. We believe such a theory
is plausible because these forces are all governed by the same principle: the
gauge principle. In fact, we shall see that gravity also obeys this rule, which
begs the question, ”Are the four known forces all aspects of some single
unified force?” Well, nobody knows, nor is it the purpose of this paper to
chase that dream. Rather, we will investigate some of the historical devel-
opments which transformed this fundamental notion from a triviality into a
cornerstone of physics.
Between the times of Maxwell and Salam, there were a number of other
advances in unification. Many theories, such as Hermann Weyl’s attempt
to unify gravity and electricity in 1918, had to be abandoned. However, in
Weyl’s case, we shall see that a slight modification of his original proposal
1
forms the foundation of what is now known as gauge theory. The generaliza-
tion of this concept, discovered by Yang and Mills, is the framework which
explains both nuclear forces. We will restrict our attention, to the gravita-
tional and electromagnetic (EM) forces. In particular, how they are both
derivable from the gauge principle.
2.1 Electromagnetism
The gauge principle was first recognized in electromagnetism, but in a rather
trivial sense. We require the following definitions:
2
With this notation, and appropriate units, Maxwell’s equations are com-
pactly written:
leaves Fµν unchanged, for any differentiable scalar function α(x). Thus,
Maxwell’s equations are unaltered by adding a gradient. Such a transforma-
tion is now termed a gauge transformation, for reasons that will become more
clear. Furthermore, for fixed x it is easy to show that the transformations
(2) form a commutative (abelian) group, with a single continuous parameter
α(x).
For a long time, the EM potential Aµ was thought of only a mathematical
tool for simplifying calculations, only the field Fµν had any physical reality.
So this gauge freedom, that is the ability to add a gradient onto the potential,
was originally considered useful, but unphysical.
3
Since we write the scalar product between two vectors (at the same point) as
u · v = gµν uµ v ν , then a vector’s squared length is given by |v|2 = (gµν v µ v ν ) =
(vν v ν ). We see that this length is invariant under parallel transport
dv 2 = d(vν v ν ) (5)
= dvν v ν + vν dv ν
= vλ Γλµν dxµ v ν − vν v µ Γνµλ dxλ
= 0.
In order for derivatives to remain co-ordinate invariant, that is gauge in-
variant, we must modify the partial derivative operator. Otherwise, a change
of co-ordinates, from primed to unprimed, for a covector’s partial derivative
yields:
0 0 0
∂µ0 v ν = (∂µ0 xµ ∂ν xν ∂µ + ∂µ0 xµ ∂µν v ν )v ν . (6)
The first term is fine, however, the second term is not a tensorial transfor-
mation. So, instead we use the covariant derivative, ∇µ (Γ), defined as
(∇µ )λν = δνλ ∂µ + Γλµν
(where δβα is the Kronecker delta) so that
0 0
∇ µ0 v ν = ∂µ0 xµ ∂ν xν ∇µ v ν . (7)
In other words, the covariant derivative transforms tensorialy. We will
see that covariant derivatives are at the heart of gauge theory; through them,
global invariance is preserved locally. The final essential geometric ingredient
for GR is the Riemann curvature tensor, which can be expressed in terms of
the connection, or the covariant derivative, as
λ
Rσµν = ∂µ Γλσν − ∂ν Γλσµ + Γλαµ Γασν − Γλαν Γασµ
= [∇σ , ∇µ ]λν .
Note that the second definition highlights the non-commutivity of parallel
transport, which tells us about the curvature of spacetime. We will write the
λ
contracted Riemann tensor Rµνλ ≡ Rµν , and the Ricci scalar R ≡ Rνν . Now
we can write Einstein’s field equations for gravitational interactions:
1
Rµν − Rgµν = Tµν (8)
2
where Tµν is the (symmetric) stress-energy tensor.
4
3 Weyl’s Unified Theory
As powerful and profound as Einstein’s gravitational theory was, many felt
it was only the beginning. To describe both known forces (the nuclear forces
were not yet discovered), the EM field tensor had to be put in by hand.
Many, including Einstein himself, sought a unified theory to explain both
phenomena, preferably in a geometric fashion like GR.
The first attempt to generalize GR to encompass EM, was proposed by
Weyl three years later. Unhappy with Riemannian geometry, Weyl devel-
oped his purely infinitesimal geometry which did not allow comparison at a
distance.
Remark 1. This was the first deliberate application of the gauge principle.
In Riemann’s geometry, the metric is fixed up to a global scale factor. Weyl’s
idea was to make that scale a local property of the metric.
5
Remark 2. The term gauge was introduced into mathematics and physics
by Weyl during this period. Until now, its use in this paper has been purely
from a modern perspective.
That is, the scalar products at P and P 0 are not equal, rather they are
proportional. The factor of proportionality 1 + dφ, which is infinitesimally
close to unity, distinguishes this geometry from Riemann’s. By expanding
(11) up to linear differential terms, we have
where we have used the substitution (10) for duα and dv β , and then used the
metric to lower the indices of { }. For the above relation to hold, for any
vectors u and v, we require that
6
where we have used the chain rule in the second line. Thus, dφ is a linear
differential form: dφ = φν dxν . Plugging this into the above, we find
7
EM potential, Aµ . Weyl postulated that of all the possible conformal connec-
tions in the equivalence class, only one had any relation to physics. Specifi-
cally, the connection where φν = γe Aν , where e is the electron charge and γ
is an undetermined constant.
This is a most remarkable result. By allowing scales to vary between
points on a manifold, Weyl could to formulate a unified theory of EM and
gravity. The gravitational fields are encoded in the metric tensor, and EM
fields are derived from the length connection form. What is perhaps even
more remarkable is this theory sprung forth from a purely mathematical
concept: Weyl’s infinitesimal geometry. GR, however, was derived from an
entirely physical fact: the equivalence of inertial and gravitational mass.
With this choice of the length connection φ, the Weyl connection is writ-
ten
e
(λαβ ) = Γλαβ + (δαλ Aβ + δβλ Aα + gαβ g νλ Aν ).
2γ
It depends on both the gravitational fields, gαβ , and the EM potential Aµ .
We can form all the objects from Riemannian geometry, such as the covariant
derivative and curvature tensors, by replacing the Christoffel connection with
Weyl’s.
Rather than develop Weyl’s geometry further, let’s return to equation
(11). Suppose a vector, v µ at P has the (squared) length l = gµν v µ v ν , then
under parallel transport to P 0 , this length changes by
e
dl = ldφ = l Aµ dxµ . (18)
γ
If instead, this vector is (parallel) transported along the path C to some
distant point Q, then upon integrating (18), we find
e
R
Aµ (x)dxµ
l = lo e γ C , (19)
where lo is l at P . So, a vector’s length is, in general, path dependant. At
this point, a clever reader might raise the following objection. Suppose we
have two identical clocks, at P . Let l be the length of a time-like vector cor-
responding to some unit of time. Now, transport the two clocks on different
paths C1 and C2 , which both end at Q. Let l1 and l2 denote the new values
of l given by each clock, at the point Q. Then, by Stokes theorem, l1 = l2 if
and only if
I Z
Aµ (x)dxµ = Fµν (x)dxµ ∧ dxν = 0 (20)
C1 −C2 D
8
where D = int(C1 −C2 ) and Fµν = ∂µ Aν −∂ν Aµ is the EM field tensor. Thus,
in the presence of an EM field, the two clock rates will differ. As Einstein
pointed out, the frequency of the spectral lines of atomic clocks would depend
on the location, both past and present, of the atom. However, we know the
atomic spectral lines to be quite definite, and independent of position.
Einstein was able to refute Weyl’s theory, with this simple physical ar-
gument. Weyl probably wished he had not sent his paper to Einstein to be
published, since Einstein included this negation as a postscript. He, nonethe-
less, admired Weyl as a brilliant mathematician, and was greatly impressed
by his novel geometric ideas. Weyl, on the other hand, was not convinced
and continued to develop his true infinitesimal geometry. He thought,
It would be remarkable if in Nature there was realized instead
an illogical quasi-infinitesimal geometry, with an electromagnetic
field attached to it.
Weyl’s gauge theory was paid little heed during the next decade. With the
coming of the Quantum era, attention moved to the microscopic regime.
9
wave function, by acting on it with linear operators. Following Schrödinger’s
lead, let’s replace the momentum variable p by the momentum operator
h̄ ~
i
∇ and the parameter E by the energy operator ih̄∂t , where h̄ is Planck’s
constant, and ∇~ is the usual 3-dimensional gradient operator. Applying both
sides to the wave function ψ we have found the Schrödinger equation
h̄ ~ 2
ih̄∂t ψ(t, x) = − ∇ ψ(t, x) + V (t, x)ψ(t, x). (22)
2m
As a simple example, consider the plane wave
which has momentum p = h̄k and energy E = h̄ω. It is not difficult to see
p2 .
that it solves the free Schrödinger equation, that is V ≡ 0, and E = 2m
The probability of finding a particle in a given volume Ω is given by the
formula
Z
Pψ (Ω) = |ψ(t, x)|2 d3 x. (24)
Ω
|ψ|2 has the natural definition then of a probability density. Wave functions
are always normalized so that the integral of the probability density over
all space equals unity. However, this does not determine ψ uniquely. For
example, ψ 0 = eiα ψ, for some real number α, (look familiar??) has the same
probability density as ψ. Thus, the states ψ and ψ 0 are equivalent.
10
The Schrödinger equation should remain unchanged, since both states are
equivalent. However, the transformed equation becomes (suppressing (t, x)
dependance)
h̄2 ~ 2 iα
ih̄∂t (eiα ψ) = − ∇ (e ψ). (26)
2m
Clearly, when the derivatives act on the exponentials, new terms will be
introduced. If we consider any single spacetime derivative, then we have
Dµ ψ → eiα Dµ ψ, (28)
where Aµ (x) is some covector field. Then under the local phase rotation,
(30) tell us that this covariant derivative transforms as
11
Which implies
A0µ = Aµ − ∂µ α. (33)
h̄2 ~
ih̄(∂t + iA0 )ψ(t, x) = − (∇ + iA)2 ψ(t, x) (34)
2m
is invariant under the simultaneous local transformations
F = mẍ = ṗ (36)
( ˙ signifies time derivatives) there must exist a force which performs these
changes. That force is none other than EM. The equation
h̄2 ~ iq
ih̄∂t ψ(t, x) = − (∇ − A)2 ψ(t, x) + qφ(t, x)ψ(t, x) (37)
2m h̄
where q is a particle’s charge, is empirically known to govern the motion of
a charged particle in an arbitrary EM field. This is equivalent to equation
(35) if we multiply α and Aµ by h̄q . The gauge transformations now read
q
ψ(t, x) → e h̄ α(t,x) ψ(t, x) and Aµ (t, x) → Aµ (t, x) + ∂µ α(t, x).
iq
(38)
h̄
So, local phase invariance introduces EM interactions.
Interestingly, the above transformations are identical to Weyl’s earlier
gauge transformations after a) replacing the metric gµν by the wave function
ψ, and b) setting the undetermined constant γ to h̄i . The first substitution
tells us that EM is a phenomenon that accompanies matter fields, and not,
as Weyl thought, the spacetime metric. Changing the constant γ from real
12
to imaginary, takes Weyl’s conformal factor from the positive real axis to the
unit circle in the complex plane. Weyl’s non-physical path dependant vector
lengths become the well proven path dependant matter-wave phases.
Historically, these correlations were first pointed out by Schödinger and
London in the 1920’s, though only in a rather tentative manner. Then, in
1929, Weyl published the paper Electron and Gravitation which introduced
many now fundamental concepts, including a derivation of EM from the
gauge principle. Before its release, Weyl published a short summary to which
Pauli, upset by the mathematician’s intrusion into physics, replied,
However, after reading the whole article, Pauli wrote back saying,
13
to the Equivalence Principle, we may always choose our reference frame, at
each point in spacetime, as a Minkowskian one, so that there appears to be
no gravitational force! By re-expressing the Minkowski basis {xm } in terms
of a general one {xµ }, we see the effect of gravity. By formulating our physics
in the {xm } basis, they are independent of any relabelling {xµ }. This is the
gauge invariance of gravity. (Proceeding in this manner is particularly useful
for constructing spinors in curvilinear co-ordinates.)
The effects of gravity are contained in the changes of {xm } from point to
point. Expressing this basis in terms of the general one, we have
dxm = hm
µ (x)dx
µ
where hm m
µ (x) ≡ ∂µ x .
hm λ λ m λ
µ Dν vm = ∂ν vµ + Γµν vλ with Γµν ≡ hµ Dν hm . (41)
∇ν vµ ≡ ∂ν vµ + Γλµν vλ .
If there is no torsion (Γλµν = Γλνµ ), then Γλµν is the Christoffel connection. One
important consequence of this approach is to free the covariant derivative
14
from the notion of parallel displacement. Instead the more fundamental
gauge principle is used.
Again, this section is by no means a rigorous derivation of GR from
the gauge principle, rather, it has been presented to highlight the analogies
between EM and GR as gauge theories. The inclined reader should consult
[3] for a complete account. Also suggested is Utiyama’s paper, Invariant
Theoretical Interpretation of Interaction, translated in [2], where a general
method for constructing gauge invariant interactions is developed, with EM,
Yang-Mills and GR as worked examples. Though published after Yang and
Mills historic paper, it was in fact, written one year prior.
Some known aspects of the gauge group structures of EM and gravitation will
be presented to highlight the differences between the two as gauge theories.
Recall the gauge principle states that ”systems invariant under a global
group of transformations, should remain invariant when that group is con-
sidered locally”. For EM, that group is, of course, the phase transformation
eiα(t,x) . Each element may be represented as a point on the unit circle in the
complex plane, formally this group is known as U (1). It has a commutative
structure and a compact topology. Also, it depends on a single parameter
α(t, x). On the other hand, GR is invariant under change of co-ordinates;
that is rotations and translations in spacetime. These transformations form
a group, as well, called the P oincaré group (PG). There are 10 indepen-
dent parameters: six for the Lorentz transformations and four for spacetime
translations. Although translations are commutative, the (generalized) rota-
tions are not, thus PG has a non-commutative structure. Furthermore, PG
is non-compact since translations are unbounded.
Some other notable distinctions between the forces are the following. In
EM, the nature of interactions is determined by the sign of the charge q,
while for gravity all matter is attractive. Apart from the charge, all EM
15
objects arise from the 1-form (also called the connection) Aµ , while in GR it
is not the connection Γλµν but the 2-form gµν which plays the primary role.
The most remarkable fact, though, is the similarity between the two
forces. The gauge principle is truly fundamental to the nature of each force.
6 Conclusions
Perhaps the most important result is that we have found the necessary con-
dition to make symmetries local. It appeared in GR, and again for EM: the
covariant derivative. The gauge principle, presented near the outset of the
paper, should include the provision:
These fields correspond to the four known force fields. By generalizing the
procedure of section 4.2, we may develop gauge invariant theories for the
nuclear forces as well.
Clearly, the gauge principle’s role in modern physics is key. From it,
we can determine the nature of the force fields, as well as their interactions
with matter. It serves as the base upon which unification theories are built.
Whether a unified theory for all interactions will ever be developed is hard
to say. That the gauge principal will play a central role in it is hard to deny.
7 References
1. C. Cohen-Tannooudji, et.al, Quantum Mechanics. John Wiley and Sons,
Toronto, 1977.
2. L. O’Raifeartaigh, The Dawning of Gauge Theory. Princeton University
Press, New Jersey, 1997.
3. P. Rammond, Field Theory: A Modern Primer. Addison-Wesley Publish-
ing, Don Mills, 1989.
4. E. Scholz, Hermann Weyl’s Raum-Zeit-Materie and a General Introduc-
tion to His Work. Birkhäuser Verlag, Boston, 2001.
5. H. Weyl, Space Time Matter. Dover Publications, New York, 1950.
16